Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Enable MLIR in CI #2134

Closed
wants to merge 23 commits into from
Closed

Enable MLIR in CI #2134

wants to merge 23 commits into from

Conversation

causten
Copy link
Collaborator

@causten causten commented Aug 30, 2023 •
edited
Loading

Reenables MLIR to CI and to the builds by default in general. Still requires MIGRAPHX_ENABLE_MLIR=1 to execute MLIR code. Tested with SLES and Ubuntu. MLIR commit hash now supports INT8

@causten causten requested review from pfultz2 and umangyadav August 30, 2023 03:11
@causten causten self-assigned this Aug 30, 2023
@codecov
Copy link

codecov bot commented Aug 30, 2023 •
edited
Loading

Codecov Report

Merging #2134 (29b86ac) into develop (4bce712) will increase coverage by 0.08%.
Report is 23 commits behind head on develop.
The diff coverage is 99.17%.

❗ Current head 29b86ac differs from pull request most recent head d64562e. Consider uploading reports for the commit d64562e to get more accurate results

@@             Coverage Diff             @@
##           develop    #2134      +/-   ##
===========================================
+ Coverage    91.40%   91.49%   +0.08%     
===========================================
  Files          422      424       +2     
  Lines        15727    15881     +154     
===========================================
+ Hits         14376    14530     +154     
  Misses        1351     1351
Files Changed Coverage Δ
src/include/migraphx/matcher.hpp 97.14% <ø> (ø)
src/include/migraphx/op/contiguous.hpp 100.00% <ø> (ø)
src/include/migraphx/op/convolution.hpp 96.51% <ø> (ø)
src/include/migraphx/op/scatter.hpp 100.00% <ø> (ø)
src/include/migraphx/shape.hpp 88.88% <ø> (ø)
src/include/migraphx/simplify_reshapes.hpp 100.00% <ø> (ø)
src/targets/ref/lowering.cpp 96.90% <ø> (ø)
src/include/migraphx/op/pooling.hpp 94.92% <97.43%> (+1.12%) ⬆️
src/common_dims.cpp 98.33% <98.33%> (ø)
src/fuse_pointwise.cpp 98.41% <100.00%> (+0.54%) ⬆️
... and 21 more

@migraphx-bot
Copy link
Collaborator

migraphx-bot commented Aug 30, 2023 •
edited
Loading

Test Batch Rate new
d64562		 Rate old
28fad2		 Diff Compare
torchvision-resnet50 64 2,282.85 2,282.65 0.01% ✅
torchvision-resnet50_fp16 64 5,358.16 5,356.46 0.03% ✅
torchvision-densenet121 32 1,833.03 1,829.72 0.18% ✅
torchvision-densenet121_fp16 32 3,386.66 3,385.91 0.02% ✅
torchvision-inceptionv3 32 1,338.54 1,342.12 -0.27% ✅
torchvision-inceptionv3_fp16 32 2,579.15 2,584.17 -0.19% ✅
cadene-inceptionv4 16 681.36 680.65 0.10% ✅
cadene-resnext64x4 16 590.54 591.36 -0.14% ✅
slim-mobilenet 64 7,216.98 7,223.75 -0.09% ✅
slim-nasnetalarge 64 236.87 237.11 -0.10% ✅
slim-resnet50v2 64 2,527.17 2,527.89 -0.03% ✅
bert-mrpc-onnx 8 720.88 721.16 -0.04% ✅
bert-mrpc-tf 1 391.86 389.70 0.55% ✅
pytorch-examples-wlang-gru 1 304.35 301.30 1.01% ✅
pytorch-examples-wlang-lstm 1 308.47 315.88 -2.35% ✅
torchvision-resnet50_1 1 553.40 555.32 -0.35% ✅
torchvision-inceptionv3_1 1 307.05 308.88 -0.59% ✅
cadene-dpn92_1 1 365.45 352.69 3.62% 🔆
cadene-resnext101_1 1 220.26 220.74 -0.22% ✅
slim-vgg16_1 1 224.58 224.38 0.09% ✅
slim-mobilenet_1 1 1,500.68 1,495.73 0.33% ✅
slim-inceptionv4_1 1 219.90 220.61 -0.32% ✅
onnx-taau-downsample 1 322.82 322.60 0.07% ✅
dlrm-criteoterabyte 1 21.68 21.68 -0.03% ✅
dlrm-criteoterabyte_fp16 1 40.55 40.45 0.24% ✅
agentmodel 1 5,870.20 5,812.23 1.00% ✅
unet_fp16 2 55.01 55.07 -0.10% ✅

Check results before merge 🔆

@migraphx-bot
Copy link
Collaborator


    :white_check_mark:bert-mrpc-onnx: PASSED: MIGraphX meets tolerance

    :white_check_mark:bert-mrpc-tf: PASSED: MIGraphX meets tolerance

    :white_check_mark:pytorch-examples-wlang-gru: PASSED: MIGraphX meets tolerance

    :white_check_mark:pytorch-examples-wlang-lstm: PASSED: MIGraphX meets tolerance

    :white_check_mark:torchvision-resnet50_1: PASSED: MIGraphX meets tolerance

🔴torchvision-inceptionv3_1: FAILED: MIGraphX is not within tolerance - check verbose output


🔴cadene-dpn92_1: FAILED: MIGraphX is not within tolerance - check verbose output


    :white_check_mark:cadene-resnext101_1: PASSED: MIGraphX meets tolerance

    :white_check_mark:slim-vgg16_1: PASSED: MIGraphX meets tolerance

    :white_check_mark:slim-mobilenet_1: PASSED: MIGraphX meets tolerance

🔴slim-inceptionv4_1: FAILED: MIGraphX is not within tolerance - check verbose output


    :white_check_mark:dlrm-criteoterabyte: PASSED: MIGraphX meets tolerance

    :white_check_mark:agentmodel: PASSED: MIGraphX meets tolerance

    :white_check_mark:unet: PASSED: MIGraphX meets tolerance

@pfultz2
Copy link
Collaborator

pfultz2 commented Aug 30, 2023

We still need fixes for ubuntu 22.04.

@causten
Copy link
Collaborator Author

causten commented Aug 30, 2023

Gated until 22.04 docker is functioning

@causten causten mentioned this pull request Aug 30, 2023
Open
pfultz2 and others added 19 commits August 31, 2023 10:42
@pfultz2
Fix mlir compile error on ubuntu 22.04 (#2138)
808a57f
@causten
update sles docker to rocm5.6 (#2133)
ac96370
@ravil-mobile
Fixed printing to stdout when MIGRAPHX_TRACE_MLIR is enabled (#2148)
1403517
@umangyadav
Skip "find_split_transpose" if split has more than one outputs (#2126)
53c406b
@kahmed10
rename memoryType to type (#2139)
d4cfdb3
@ahsan-ca
Move ref tests under a separate directory and split them based on the…
a0894c2 
… type of operation (#2106)

Re-organizes ref tests to improve the speed of compilation
@pfultz2
Fuse pointwise modules across reshapes (#1940)
1d8840b
@ravil-mobile
Fixed writing the input files for rocMLIR's tuningRunner.py (#2143)
bded094
@ravil-mobile
Fixed split_string from stringutils (#2160)
20f6ed9
@lakhinderwalia
Move operation for memory performance + misc changes for cpu performa…
b00489b 
…nce (#2130)

Reduce memory footprint by std::move of temporary (potentially very large) containers.
Minor cleanup for performance optimization: e.g. of Index() calculation -- which can get repeated millions of times in large tensors/vectors in a single Visit.
@CharlieL7
Dynamic allocate (#2079)
ede8bfa 
Makes a version of allocate that takes in dimensions and allocates a buffer
Going to create a simplify_dynamic_ops compiler pass that will use the use_shape_attr flag
The ONNX op ConstantOfShape needs the buffer to be filled with a specific value, so going to make another PR for that and a fill operator
@tvukovic-amd
Fixed test float equal for Windows (#2153)
37787e0 
Added case to skip test_limits<int,long> while running tests on Windows since int and long have same min and max values
@tvukovic-amd
Fixed stringutils test (#2154)
dfd95c5
@pfultz2
Add environment variable to enable mlir exhaustive tuning configs (#2103
22d1866 
)
@github-actions
Update onnxruntime main ae74a517b62baa6d973e46b5b51ac9a640512c46 (#2163)
72b691a
@ravil-mobile
[Re-Opened] Added support for standalone mlir-conv (used to be #2110) (…
ea97ce5 
…#2142)
@pfultz2
Ensure same layout is used for miopen convolution (#2102)
64b306a
@krzysz00
Bump MLIR commit pointer to add align-tiling fix (#2170)
28fad2b 
In addition, since the bug in the sigmoid test is resolved, also uncomment that temporary disable.
@pfultz2
Disable unsafe buffer usage warning when its available (#2168)
e05c94b
@causten
Copy link
Collaborator Author

causten commented Sep 25, 2023

Messed up the rebase, using a new branch

@causten causten closed this Sep 25, 2023
@causten causten deleted the add_mlir branch January 5, 2024 19:58
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@umangyadav umangyadav umangyadav approved these changes

@pfultz2 pfultz2 Awaiting requested review from pfultz2

Assignees

@causten causten

Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.